Acute myelogenous leukemia (AML) accounts for approximately 20 percent of cases of childhood leukemias. In contrast to lymphoblastic leukemias, the cure rate for childhood AML is still less than 50 percent. We propose to apply the powerful new technology of gene expression profiling to AML specimens from more than 600 children with de novo AML treated on the Pediatric Oncology Group (POG) Study #9421. Our hypothesis is that global gene expression profiles will provide new insights into genetic determinants of response to therapy and clinical behavior of childhood AML. The specific aims of this proposal are: (1) Gene expression profiling of childhood de novo AML specimens from patients enrolled in POG Study #9421. A total of 621 banked specimens are available. The microarrays for this project will contain 48,000 human genes. The profiles of gene expression of childhood AML will be compared to our existing databases including adult AML, lymphomas, and several types of carcinomas. In addition, gene expression clusters will be analyzed with reference to histopathological classification and cytogenetic abnormalities. It is likely that this approach will reveal clusters of genes which will re-define subclasses of childhood AML. (2) Prognostic significance of genes involved in drug resistance and apoptotic pathways. Modulation of drug resistance by high dose cyclosporine and high dose cytarabine were the two therapeutic questions addressed by #9421, resulting in 4 study cohorts. Although early in follow-up, there is a trend to increased survival with high dose cyclosporine. Drug transporters and other genes known to be involved in therapeutic responses will be examined for prognostic importance in each treatment cohort. We hypothesize that MDR1 will lose prognostic importance in the cyclosporine cohorts. The relationship of individual and clustered gene expression to complete remission, duration of remission, and survival will be assessed for the entire set of genes on the arrays. (3) Gene expression profiling of specimens from patients with relapsed pediatric AML enrolled in POG #9421 and #9720. Serial specimens at diagnosis and relapse are available in at least 39 cases, providing a great opportunity to identify genes which may be selected or induced by therapy. The outcome of these studies may guide the next generation of biologic and therapeutic studies in childhood AML, and make a significant contribution to the diagnosis, prognosis and development of rational therapeutics in this disease.